Michelle Anditio

I've built UMKMentor, an AI business analysis platform for Indonesian MSMEs using React and FastAPI. Its XGBoost product-success model achieved 0.760 Test AUC and 68% accuracy, while its IndoBERT sentiment model reached 86.79% accuracy and an 87.58% weighted F1-score.

I also built Promise Yura, an AI-powered reminder web application with React, Firebase, conversational AI, authentication, and real-time storage during a one-month hackathon completed in 2–3 weeks.

My projects span TensorFlow image classification at 95% accuracy and an IoT baby-room monitoring dashboard that connects ESP32 sensors with Firebase for real-time safety status, motion sensing, and LED indicators.

I enjoy taking products from UI/UX design in Figma through full web development, including collaborative work on the Zero Waste website and its education, community, directory, and Google Maps features.